Image Modeling: A Mathematical Framework for Segmentation and Object Detection.
Abstract
Decision rules for segmenting scenes and for detecting the presence of distinguished objects in digital images can be based on classical principles of statistical principles of statistical inference if appropriate mathematical image models are developed for observable pictures. The main goal of this research was to devise and analyze alternation image models for digitized FLIR images. The work has been done in close cooperation with the Advanced Modeling Team of the U.S. Army Night Vision and Electro-Optics Laboratory, Ft. Belvoir, Virginia. This report concentrates of hierarchical Markov Random Field models and their application to restoration and segmentation of FLIR images. Keywords: Image processing; Bayesian methods; Infrared images. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 20, 1987
- Accession Number
- ADA179904
Entities
People
- Donald E. Mcclure
- Donald Geman
- Stuart Geman
- Ulf Grenander